structural and functional similarities between the human ... › content › joces › 110 › 16...

1893Journal of Cell Science 110, 1893-1906 (1997)Printed in Great Britain © The Company of Biologists Limited 1997JCS9655

Structural and functional similarities between the human cytoskeletal protein

zyxin and the ActA protein of Listeria monocytogenes

Roy M. Golsteyn 1, Mary C. Beckerle 2, Tom Koay 1 and Evelyne Friederich 1,*1Morphogenèse et Signalisation Cellulaires, Centre National de la Recherche Scientifique, UMR 144, Institut Curie, 26 rue d’Ulm,Paris 75248 Cedex 05, France2Biology Department, University of Utah, Salt Lake City, Utah, 84112, USA

*Author for correspondence (e-mail: [email protected])

The intracellular bacterial parasite Listeria monocytogenesproduces ActA protein at its surface to facilitate thelocalized assembly of actin-filled comets that are requiredfor movement. The organization of actin in Listeria cometsshows striking similarity to the organization of actin at theplasma membrane of mammalian cells. Therefore weexamined the possibility that an ActA-like protein ispresent in mammalian cells. By using antibodies directedagainst ActA, we identified zyxin as an ActA related proteinin a number of cell types. We compared the functions ofActA and zyxin by transient expression of variants taggedwith an inner plasma membrane localization sequence (aCAAX box). Targeting of the proline rich domain of zyxin

to the plasma membrane disrupts the actin cytoskeletonand cell shape in a manner similar to that which occurswith membrane-targeted ActA sequences. A chimericprotein composed of the N-terminal domain of ActA fusedto the N-terminal and central domains of zyxin induced afull ActA response in cells. Furthermore, zyxin and ActAexhibit common protein partners in vitro. On the basis ofthe shared properties of zyxin and ActA, we propose thatzyxin enhances actin organizing activity in mammaliancells.

Key words: ActA, Actin, Cell motility, Listeria, Zyxin

SUMMARY

d)the of

al.,g)l.,

er- toing

g isel-

-m

endiot,iz-

tintherel.,

INTRODUCTION

The actin cytoskeleton is important for cell shape, polarity amovement. Within cells actin is organized into arrays of acfilaments such as those found in microvilli, lamellipodiextensions, the contractile ring, and stress fibers. The formaof these structures is regulated by numerous actin interacproteins that in turn are regulated by protein phosphorylatismall GTPases or by signaling molecules such as phphatidylinositols or calcium. Often, different types of actin-ricstructures can be found in a single cell. These structuresassembled at specific sites in a cell and at specific timHowever, little is understood about how the cell is able restrict the assembly of filamentous actin to discrete regioAn important task in cell biology is to identify proteins thainfluence the spatial and temporal control of actin filameassembly in different cell types.

One region of a eucaryotic cell that is specialized fregulated actin assembly is the leading edge. Here the orization of actin is distinct from that in other regions of the c(Forscher and Smith, 1988; Stossel, 1993; Lauffenburger Horwitz, 1996). For example, the leading edge has the grearate of actin turnover relative to other actin rich structur(Wang, 1985; Okabe and Hirokawa, 1989; Theriot aMitchison, 1991). Another area that displays prominent actmembrane interactions is the adhesion plaque, where thecontacts the extracellular matrix. At these sites, bundles

ndtinaltiontingon,os-h arees.tons.tnt

organ-ellandtestesndin- cell of

actin filaments are oriented with their fast growing (barbeends toward the plasma membrane. Actin polymerization at plasma membrane may originate by generation of new sitesformation (nucleation) as observed in leukocytes (Cano et 1991) or by elongation of pre-existing filaments (uncappinas described in platelets (Hartwig et al., 1995; Barkalow et a1996; Schafer et al., 1996). Furthermore, the rate of polymization can be influenced by recruitment of actin monomersthe plasma membrane and by small actin monomer bindproteins such as profilin and thymosinβ4 (Pantaloni andCarlier, 1993). Identifying the precise roles of actin interactinproteins in actin polymerization at the plasma membranedifficult because the dynamic properties of the actin cytosketon are disrupted by most biochemical techniques.

A promising approach to study the control of actin polymerization in vivo comes from analysis of the bacteriuListeria monocytogenes(Tilney and Portnoy, 1989). L. mono-cytogenesis one of several micro-organisms that can invadeucaryotic cells and use host cell actin during infection aspread (Sheehan et al., 1994; Cudmore et al., 1995; Ther1995). There are striking similarities between actin organation within L. monocytogenescomets and the lamellipodiumof mammalian cells. Comets are composed of short acfilaments in which the barbed ends are oriented toward bacteria and many host cell actin interacting proteins arecruited to the actin comet (Small et al., 1978; Dabiri et a1990; Tilney et al., 1992a). L. monocytogenesproduces ActA,

1894 R. M. Golsteyn and others

a surface protein that is essential to recruit actin and form actinfilled comets (Domannet al., 1992; Kocks et al., 1992; Tilneyet al., 1992b). By expression of the actA gene in unrelatedbacterial species and by attachment of ActA to bacterial orparticle surfaces, it has been shown that ActA is likely to bethe sole bacterial protein required to organize host cell actin(Kockset al., 1995; Smithet al., 1995). In addition, expressionand targeted localization of ActA in mammalian cells bytransient transfection of actA cDNA (independent of bacterialinfection) causes the recruitment of host cell actin to the sitesof ActA enrichment (Friederich et al., 1995; Pistoret al., 1995).

ActA exhibits at least two major domains that are essentialfor its activity. An N-terminal domain of 234 amino acids isneeded for actin filament formation as revealed by analysis ofthe behavior of L. monocytogenescarrying mutations in theactA gene (Lasaet al., 1995, 1997) or by expression of cDNAsencoding truncated ActA by transient transfection in tissueculture cells (Friederich et al., 1995; Pistor et al., 1995). Theprecise function of the N-terminal region of ActA is not known,but it may be responsible for actin nucleation activity associ-ated with the surface of L. monocytogenes(Tilney et al., 1990,1992b). The remainder of ActA (amino acids 235-584contains a proline rich region whose precise function alsoremains to be clearly defined. By assays similar to those usedto probe the function of the N-terminal domain, it appears thatthe domain rich in prolines greatly stimulates comet formationand bacterial movement (Southwick and Purich, 1994;Friederich et al., 1995; Pistor et al., 1995; Smithet al., 1996).Interestingly, an ActA-like molecule, IactA, has been identi-fied in a related species of bacteria, Listeria ivanovii. AlthoughIactA and ActA proteins share only 30-40% sequence identity,they have similar actin recruitment properties (Chakraborty etal., 1995; Gouinet al., 1995; Kreft et al., 1995; Gerstel et al.,1996).

The ActA protein interacts with host cell factors that are pos-tulated to play a role in actin filament assembly. For example,ActA interacts with proteins in the Ena/VASP family that bindprofilin and are localized at certain sites of actin-membraneinteraction (Chakraborty et al., 1995; Gertler et al., 1996). Inaddition, ActA binds to a protein complex (the ARP complex)which contains two actin-related proteins and has been impli-cated in actin filament nucleation (Welch et al., 1997). Theability of ActA to interact with proteins that are normally foundin actin rich structures strongly suggests that an ActA-likeprotein may exist in eucaryotic cells. The identity of such aprotein would be an important step toward understanding actinbased cell motility in mammalian cells.

With this objective in mind we developed a variety ofmethods to characterize ActA-like proteins in mammaliancells. We raised antibodies directed against ActA for use astools to identify human proteins that have common epitopeswith ActA. By this approach we have identified the cytoskele-tal protein, zyxin (Beckerle, 1986; Crawford and Beckerle,1991). We have also examined the functional relationshipsbetween zyxin and ActA. We used a transient transfectionassay in which zyxin and ActA were expressed in fusion witha peptide sequence derived from K-ras (B) to direct them tothe inner plasma membrane of cells (Friederich et al., 1995).Ectopically localized ActA and zyxin triggered a similar,dramatic reorganization of the actin cytoskeleton. In addition,biochemical studies revealed that zyxin and ActA have similar

)

binding partner preferences. The demonstration that zyxinexhibits certain ActA-like properties leads us to propose thatzyxin may have an important role in organizing the dynamicproperties of the actin cytoskeleton of eucaryotic cells.

MATERIALS AND METHODS

Generation of ActA-specifi c antibodiesGlutathione S-transferase (GST) fusion proteins encoding theprocessed form of ActA (ActA1: amino acids 1-584; numbering isaccording to Kocks et al., 1992), N terminus ActA (ActA2: aminoacids 1-234), a medial and C-terminal region including the prolinerich domain (ActA3: amino acids 235-584), and a C-terminal regionwithout the proline rich domain (ActA4: amino acids 394-584) werepurified from bacterial extracts by affinity chromatography (Smith andJohnson, 1988). Polyclonal antiserum was prepared against eachprotein by routine immunization of rabbits (Harlow and Lane, 1988).Anti-ActA1 antibodies and anti-ActA3 antibodies were purified byActA affinity chromatography as described by Harlow and Lane(1988).

Cell lines and cellsThe human cervical carcinoma HeLa cell line (ATCC CCL2), thefibroblast-like monkey kidney cell line CV-1 (ATCC CL 101), humankeratinocytes (Rogel-Gaillard et al., 1992; Robineet al., 1993), humanforeskin fibroblasts (Dr A. Rochat, Département de Biologie, EcoleNormale Supérieure, Paris, France) were grown in Dulbecco’sminimum essential medium (DMEM) supplemented with 10% fetalcalf serum (complete medium). HL60 (Collinset al., 1977) and humanmyeloma RPMI 8226 (Matsuoka et al., 1967) (gifts from Dr B.Bauvois, Institut Curie) were cultured in RPMI 1640 medium sup-plemented with 2 mM glutamine and 10% fetal calf serum. All celllines were cultured at 37°C and 10% CO2. Human platelets werereceived from Dr C. Gachet (INSERM U311, Centre Régional deTransfusion Sanguine, Strasbourg, France).

Construction of pr ocar yotic and eucar yotic e xpressionvector s containing actA or zyxin sequencesDNA manipulations were performed as described by Sambrook et al.(1989). Human zyxin cDNA (GenBank Accession number X94991)was subcloned into the EcoRI site of BlueScript KS(−). For expressionin Escherichia coli, sequences encoding all or part of actA wereinserted into the pGEX-2T vector. For expression in eucaryotic cells,actA or zyxin coding sequences were inserted into the CMV-derivedpCB6 (gift of Dr M. Roth, University of Texas) or pUHD-10-3(Gossen and Bujard, 1992) expression vectors.

To generate zyxin variants that localize to the inner face of theplasma membrane, sequences encoding the 18 carboxy-terminalamino acids of K-ras (B) were fused to sequences encoding thecarboxy-terminal amino acid of the zyxin variants as has beendescribed previously (Friederich et al., 1995). This sequence is suffi-cient to anchor proteins at the plasma membrane (Hancock et al.,1991).

To generate an epitope-tagged zyxin variant that comprised aminoacids 380 to 559 of the human zyxin, fused to the K-ras (B)-derivedsequence, two complementary oligonucleotides, 5′-ATTCCG-GATATCCCATGGCCG-3′ and 5′-GATCCGGCCATGGGGATATC-CGG-3′, were inserted to fuse the sequence encoding the VSV-Gpeptide in-frame to the 5′-end of the zyxin coding sequence. Theresulting construct encoded the zyxin variant VSV-zyxin(381-559)-CAAX and comprises 11 amino acids of the VSV-G protein fused toamino acids 381 to 559 of zyxin followed by the K-ras (B) signal.

An expression construct that encoded the ActA-zyxin chimeraActA(1-234)-zyxin(1-381)-CAAX that comprises amino acids 1-234of ActA fused to amino acids 1-381 of zyxin followed by the K-ras

1895Common properties of zyxin and ActA

t

(B) sequence was generated by standard procedures. pIC7 DNA,which encodes the N-terminal 234 amino acids of ActA, was a gift ofI. Lasa and P. Cossart, Institut Pasteur, Paris, France (Lasa et al.,1995).

In order to be able to distinguish exogenously expressed zyxinsequences from the endogenous protein, a DNA construct was preparedthat encodes the N-terminal 380 amino acids of zyxin followed by 20amino acids that form an epitope for the 9E10 anti-myc antibody(VAACNMEQKLISEEDLNMNS); in the expressed protein, the mycepitope is followed by the 18 carboxy-terminal amino acids of K-ras(B) (Evanet al., 1985; Schmidt-Zachmann and Nigg, 1993). Sequenceswere confirmed by double stranded cDNA sequencing.

Transient cDNA e xpression in cultured cellsHeLa or CV-1 cells were transfected using the calcium phosphateDNA precipitation method (Matthias et al., 1982). Cells wereanalyzed 24-48 hours after the addition of DNA. In experiments withthe pUHD-10-3 vector 2.5 µg of this plasmid which contains a CMVminimal promotor were co-transfected with 2.5 µg of pUHD-15-1plasmid which encodes the tTAs transactivator (Gossen and Bujard,1992). We used this system that was initially constructed for tetra-cycline-dependent inducible expression because of its higher trans-fection efficiency when compared to the pCB6 vector.

Preparation of cell e xtracts and imm unob lottingEucaryotic cell extracts were prepared as previously described(Friederich et al., 1989). E. coli bacterial extracts were prepared bycollecting a volume of cell suspension (OD600 approximately 0.8),centrifuging at 10,000 g for 2 minutes and resuspending the pellet ina volume of SDS-sample buffer equivalent to the original suspensionvolume. Extracts of Listeria ivanovii were gifts from E. Gouin and P.Cossart of the Pasteur Institute, Paris (Gouin et al., 1995). For analysisof proteins from transfected cells, proteins from cell lysates wereseparated by sodium dodecyl sulphate (SDS)-polyacrylamide gel elec-trophoresis under reducing conditions (Laemmli, 1970). For two-dimensional gel electrophoresis (nonequilibrium pH, NEPHGE), totalplatelet extract was resolved by Bio-Rad Mini-Protean II 2-D systemas described in the manufacturer’s instructions and by O’Farrell et al.(1977). Ampholytes (pH range 3-10 and 6-8) were purchased fromServa (Heidelberg). Transfer to nitrocellulose and antibody incubationwere performed according to the method described by Burnette(1981). P5D4 monoclonal antibodies against the 11 amino-terminalamino acids of the vesicular stomatitis virus glycoprotein G (VSV G-protein), were kindly provided by T. Kreis (University of Geneva,Switzerland).

Fluorescence labeling of cellsCells were fixed with 3% paraformaldehyde, detergent permeabilizedwith 0.4% Triton X-100 and labeled as described previously(Friederich et al., 1995). For detection of vinculin, a tissue culturesupernatant of hybridoma 7F9 cells was used (gift of V. Kotelianski,Institut Curie, Paris). For detection of the VSV epitope-tagged zyxinvariant VSV-zyxin(381-559)-CAAX, monoclonal anti-VSV anti-bodies (P4D5) were used. For detection of the myc epitope-taggedzyxin variant zyxin(1-380)-myc-CAAX, 9E10 anti-myc antibody wasused (Evan et al., 1985). An anti-VASP antiserum was provided byU. Walter, Medizinische Universitätsklinik, Würzburg, Germany andanti-Mena was provided by F. Gertler, Fred Hutchinson CancerResearch Center, Seattle, WA (Gertler et al., 1996). Labeled cells wereanalyzed by confocal laser scanning microscopy (Leica TCS4D, Hei-delberg). Typically images were collected within the linear range ofdetection, 8 scans per image, and 8 to 12 images to encompass thecell volume from the basolateral to the apical surfaces. Images wereprocessed with Leica SCANware.

Overla y detection of ActA and zyxin interacting pr oteinscDNA encoding human profilin was a gift of S. Almo, Albert Einstein

College of Medicine, New York, USA. Recombinant protein wasisolated as described (Fedorov et al., 1994) and its purity wasconfirmed by SDS-PAGE. Profilin Sepharose was prepared bycoupling 5 mg of protein to 2 ml CNBr-Sepharose as described in themanufacturer’s instructions (Pharmacia). HeLa cell extracts wereincubated with profilin Sepharose (10:1 volume ratio) at 4°C for 1hour. The beads were collected by centrifugation, washed three timesin bead buffer (50 mM Tris-HCl, pH 7.4, 50 mM NaF, 200 mM NaCl,5 mM EDTA, 5 mM EGTA and 0.1% Nonidet P-40). The beads weretransferred to a new tube, mixed with 2 volumes of SDS-sample bufferand boiled. Typically, a sample equivalent to 100 µl of total cell extractwas loaded per lane for analysis by SDS-PAGE. Coomassie stainingof gels revealed that immobilized profilin retained strong actin bindingactivity.

Total HeLa cell extracts and samples containing proteins retainedby profilin-Sepharose were separated by SDS-PAGE, transferred tonitrocellulose, and blocked. GST fusion proteins containing the Nterminus of chicken zyxin (Schmeichel and Beckerle, 1994), ActA(1-584), or ActA(235-584) were added to a final concentration of 1 µg/mlto the blocking solution, and were incubated for 30 minutes with thenitrocellulose membrane. The membrane was then processed asdescribed for immunoblotting.

RESULTS

Characterization of anti-ActA antibodiesWe used an immunological approach to identify proteins inmammalian cell lines that may share sequence identity andpossibly functional similarity to the L. monocytogenesActAprotein. Antibodies were prepared against recombinant matureActA protein (ActA1: amino acids 1-584), the N-terminaldomain (ActA2: amino acids 1-234), a region including theproline rich domain and the C terminus (ActA3: amino acids235-584), and a C-terminal region without the proline richdomain (ActA4: amino acids 394-584) (Fig. 1A). These anti-bodies were purified by virtue of their affinity for recombinantActA protein.

Each affinity purified antibody was evaluated byimmunoblot analysis for its ability to recognize the ActAprotein of L. monocytogenesand the IactA protein of a relatedpathogenic bacterium, Listeria ivanovii. In Fig. 1B we showthat purified anti-ActA1 and anti-ActA3 antibodies reactedstrongly with recombinant ActA protein but did not react withproteins from total E. coli cell extracts that contained GSTprotein. The IactA protein of Listeria ivanovii is a surfaceexpressed protein that is functionally similar to ActA yet thetwo proteins share only 30-40% sequence identity (Gouin eal., 1995; Kreft et al., 1995). We reasoned that if anti-ActAantibodies recognize IactA protein, it would be likely that theantibodies would recognize functional domains in ActA-likemolecules from other species. Anti-ActA1 (full length) andanti-ActA3 (C terminus including the proline rich region) anti-bodies reacted with IactA (Fig. 1C). As previously reported,the IactA protein of strain Li 1188 migrates more rapidly thanIactA of strain Li 497 when examined by SDS-PAGE (Gouinet al., 1995). These results suggested that the predominantbacterial proteins detected by these antibodies are ActA orActA-related proteins. Anti-ActA2 and anti-ActA4 antibodiesdid not react with IactA by this method and were not used forfurther experiments in this study.

Anti-ActA antibodies recogniz e human zyxinIf the anti-ActA antibodies recognize epitopes common to


insertionmembraneproline/glutamate

sequencesignal

-29 1 235 393 584 610

ActA

Anti-ActA1

aa

Anti-ActA2Anti-ActA3Anti-ActA4

Arich

Fig. 1. Preparation and characterization of anti-ActA antibodies.(A) The different domains of full length ActA protein arehighlighted. These domains include the signal sequence (amino acid−29-1), the N-terminal domain (amino acids 1-234), the proline andglutamate rich domain (amino acids 235-393), a C-terminal domainof unknown function (amino acids 394-584) and a membraneinsertion sequence (amino acids 585-610). Numbering of aminoacids is according to Kocks et al. (1992). Recombinant ActA proteins(marked by bars) used to prepare polyclonal antibodies are listed asfollows: ActA1: amino acids 1-584; ActA2: amino acids 1-234;ActA3: amino acids 235-584; ActA4: amino acids 394-584.(B) Samples containing 5 µg of total E. coliextract (prepared afterinduction of GST leader peptide expression; lane 1) or 50 ng ofpurified GST-ActA protein (lane 2) were analyzed by SDS-PAGEand immunoblotting with affinity purified anti-ActA1 or anti-ActA3antibody. The positions of molecular mass markers (kDa) are shownon the left. (C) Surface protein extracts of Listeria ivanovii strains Li497 or Li 1188 were analyzed by SDS-PAGE and immunoblottingwith affinity purified anti-ActA1 or anti-ActA3 antibody. Thepresence of multiple forms of IactA and the difference between themobility are characteristic for the two strains. The positions ofmolecular mass markers (kDa) are shown on the left.

proteins involved in the regulation of actin filament assembly,as suggested by their ability to bind to both ActA and IactAproteins, we expected that the antibodies might react witheucaryotic proteins of similar function. To explore this possi-bility, we labeled human tissue culture cells for indirectimmunofluorescence by using affinity-purified anti-ActA1 andanti-ActA3 antibodies and examined the cells by epifluor-escence microscopy and by confocal laser scanningmicroscopy (Fig. 2). Anti-ActA3 strongly stained a number ofsmall structures in human foreskin fibroblasts (A). Todetermine if the anti-ActA3 staining corresponded to actin richstructures we stained F-actin with rhodamine-labeled phal-loidin (B). Merging the two images revealed that the anti-

ActA3 staining co-localized with the tips of stress fibers andthus may correspond to focal adhesions (C). To confirm thestaining of anti-ActA antibody at focal adhesions weperformed another double labeling experiment with anti-vinculin monoclonal antibody, a marker of focal adhesions(Fig. 2D-F). Anti-ActA3 antibody staining (D) and anti-vinculin antibody staining (E) co-distributed as shown in themerged image in F. We labeled HeLa, Caco-2 cells and a ker-atinocyte cell line with anti-ActA3 antibody and found similarstaining patterns, in addition we occasionally observed stainingof lamellipodia (not shown). Anti-ActA1 antibody also stainedfocal adhesions but the signals were much weaker than thoseseen with anti-ActA3 (data not shown). These results suggestthat epitopes present in ActA may also be present in proteinsassociated with the actin cytoskeleton.

To identify the eucaryotic protein that reacted with anti-ActA antibodies, we prepared total cell extracts fromdifferent cell lines and analyzed them by immunoblottingtechniques. Anti-ActA3 antibody reacted with a group ofclosely spaced proteins of approximately 84 kDa whenexamined by SDS-PAGE (Fig. 3A, left panel). Althoughsimilar protein amounts were tested for each cell extract, therelative amount of cross-reactive protein varied, being at thelimit of detection in RPMI 8226 cells (a B-cell line) andeasily detected in extracts from human platelets. The anti-ActA1 antibody also detected a group of closely spacedproteins that co-migrated exactly with the proteins observedwith anti-ActA3 antibody (not shown).

To confirm the relationship between the 84 kDa antigendetected by immunoblotting and the staining of actin rich struc-tures in cells analyzed by immunofluorescence, we furtherpurified the anti-ActA3 antibody by virtue of its affinity for theelectrophoretically resolved, nitrocellulose immobilized 84kDa protein. By indirect immunofluorescence, these blot-affinity purified antibodies stained actin rich structures in apattern that was indistinguishable from that obtained with theanti-ActA antibodies used in the original experiments. Thisresult suggests that the 84 kDa protein is the antigen detectedby immunofluorescence (not shown).

We compared the cell staining pattern and the molecularmass of the protein detected by anti-ActA antibodies with thoseof previously described actin associated proteins. Only zyxin,a previously described component of actin rich structures,showed similar properties to the protein detected with anti-ActA antibodies (Crawford and Beckerle, 1991). First, theproteins detected with both anti-ActA1 antibody (data notshown) and anti-ActA3 antibody co-migrated exactly withproteins detected by anti-zyxin serum (Fig. 3A, right panel).Moreover, the cell type-specific protein production pattern forzyxin is very similar to that observed with anti-ActA anti-bodies. Second, we compared the properties of the proteindetected by anti-ActA antibodies and anti-zyxin antibodies intotal cell extracts of human platelets by two-dimensional gelelectrophoresis. The proteins detected by anti-ActA3 antibodyor by anti-zyxin antibody showed a similar charge and size(Fig. 3B). Finally, to test directly if anti-ActA antibodies reactwith human zyxin protein, we produced the protein by tran-scription and translation in vitro from a zyxin cDNA (Macalmaet al., 1996). Immunoblot analysis of the reaction productsrevealed that zyxin translated in vitro is recognized by anti-ActA antibodies and co-migrates exactly with the 84 kDa

s


Fig. 2.Staining of human tissue culture cells by anti-ActA antibodies. Human foreskin fibroblasts were fixed, permeabilized, and doublystained with anti-ActA3 antibody (A,C and D,F) and phalloidin to mark actin (B,C) or with anti-vinculin monoclonal antibody (E,F). Cellswere analyzed by laser confocal scanning microscopy, and merged images are shown in C,F. Anti-ActA3 antibody strongly marked focaladhesions; here we show an example of a cell where the signal intensity varied relative to vinculin staining (F). Bars: 10 µm (C); 3 µm (F).

protein detected in cell extracts by the anti-ActA antibodies(data not shown). We conclude that anti-ActA1 and anti-ActA3antibodies detect human zyxin.

We compared the amino acid sequences of L. monocyto-genesActA with human zyxin to identify common epitopesthat have been detected by antibodies. The sequencealignment reveals that the two proteins share limited sequenceidentity, although several key related structural features areevident. The areas of sequence identity lie within the prolinerich region and C-terminal portion of ActA and the N-terminaldomain of zyxin (Fig. 4A). As in ActA, the N-terminal domainof zyxin also has a proline rich region followed a medialportion. The sequence identity between these two regions ofActA and zyxin is 23%, with the highest degree of similaritylying within short regions rich in proline and glutamate aminoacids. To examine the possibility that zyxin might sharefeatures with ActA outside of the proline-rich repeats, wefurther fractionated the anti-ActA3 antibody by selecting anti-bodies that bound to C-terminal sequences in ActA that lackedthe proline repeats (amino acids 394-584). These antibodiesstill retained the ability to react with human zyxin when testedby immunoblotting suggesting that common epitopes are alsofound outside of the proline rich region of the two molecules(not shown).

Construction of plasma membrane associatedvariants of zyxinTo test if the structural similarities detected by immunologicalapproaches are an indication of functional similarities betweenActA and zyxin, we compared the abilities of the two proteinsto affect actin organization when targeted to the cytoplasmicface of the plasma membrane. This targeted transfection assayhad previously been used to characterize the effects of ActAon the actin cytoskeleton and cell shape (Friederich et al.,1995). Five cDNA constructs were prepared for this assay; oneconstruct encoded wild-type zyxin, the remaining constructsencoded zyxin variants in frame with a CAAX motif and astretch of basic amino acids derived from K-ras (B). Thissequence has been shown to be sufficient for association ofproteins with the inner leaflet of the plasma membrane(Hancock et al., 1990). The constructs used for transfectionencoded human zyxin(1-559)-CAAX, zyxin(1-381)-CAAX,an epitope-tagged variant VSV-zyxin(381-559)-CAAX, and achimeric protein composed of the N-terminal domain of ActA(1-234) in frame with zyxin (1-381) (Fig. 4B). Transient trans-fections were performed in both HeLa cells, which have apoorly organized actin cytoskeleton, and in the simian fibro-blast-like CV-1 cells, which are flat and have a well organizedactin network (Friederich et al., 1995).


Fig. 3. Anti-ActA3 antibody detects p84, a protein whose cellexpression pattern, size, and charge are similar to zyxin. (A) Cellextracts were prepared from the following cells and cell lines: CV-1,HeLa, human keratinocytes (Kera.), human foreskin fibroblasts(HFF), human myeloma cell line (RPMI 8226), HL60, and humanplatelets (Plat.). Approximately 15 µg of protein from each extractwere analyzed by SDS-PAGE and immunoblotting with affinitypurified anti-ActA3 antibody (left panel) or anti-zyxin serum (rightpanel). The positions of molecular mass markers in kDa are shownon the left. (B) Human platelet extract was prepared in buffercontaining urea, Triton X-100, and ampholytes (pH 6-8) thenseparated by isoelectric focusing. Tube gels were analyzed by SDS-PAGE and immunoblotting with affinity purified anti-ActA3antibody (left panel) or anti-zyxin serum (right panel). The polarityof the 1st dimension is shown by (−) and (+). The positions ofmolecular mass markers (kDa) are shown on the left.

CAAX

Wild type zyxin

Zyxin(1-559)-CAAX

Zyxin(1-381)-CAAX

ActA

Zyxin

N 1 C

23% identity

N 1 C 573LIM domains

VSV-zyxin(381-559)-CAAX

CAAX

proline/glutamaterich

CAAX

CAAX

610

ActA(1-234)-zyxin(1-381)-CAAX

A

B

Fig. 4. Production of zyxin variants in transiently transfected HeLacells. (A) The structures of Listeria monocytogenesActA and humanzyxin are represented by bar diagrams that highlight the differentdomains present in each protein. ActA has a medial proline andglutamate rich domain and C-terminal domain of unknown function.Zyxin also has a proline and glutamate rich domain. (B) Thestructures of zyxin and zyxin variants tested by the targetedtransfection assay are shown. The constructs are as follows: wild-type zyxin; zyxin(1-559)-CAAX; zyxin(1-381)-CAAX, whichrepresents only the region that is similar to ActA; VSV-zyxin(381-559)-CAAX; ActA(1-234)-zyxin(1-381)-CAAX, an ActA-zyxinchimera prepared by an in frame-fusion of actagene sequenceencoding amino acids 1-234 of ActA to the 5′-end of the DNAencoding zyxin(1-381)-CAAX. (C) Cell extracts were prepared fromHeLa cells (lane 1) and from HeLa cells after transfection withcDNAs encoding, ActA-CAAX (lane 2), zyxin(1-559)-CAAX (lane3), zyxin(1-381)-CAAX (lane 4) and ActA(1-234)-zyxin(1-381)-CAAX (lane 5). Samples were analyzed by SDS-PAGEimmunoblotting with anti-zyxin serum (left panel) or affinity purifiedanti-ActA3 antibody (right panel). The positions of molecular massmarkers (kDa) are shown on the left.

The production of zyxin variants in HeLa cells wasexamined by immunoblotting with anti-ActA3, anti-zyxin, oranti-VSV antibodies. Endogenous zyxin (84 kDa) was detectedin all extracts with anti-ActA3 and anti-zyxin antibodies (Fig.4C). Anti-ActA3 antibody reacted with ActA-CAAX (97 kDa;right panel, lane 2) protein, which was prepared as a positivecontrol, and reacted with zyxin variants (right panel, lanes 3-5). In extracts of transfected cells producing zyxin(1-559)-CAAX (lane 3) a strong signal was detected at 84 kDa,enhancing the signal already observed for endogenous zyxin.In extracts of transfected cells producing zyxin(1-381)-CAAX(lane 4) a strong signal was detected at approximately 60 kDa.As expected, cells expressing the ActA-zyxin chimeraproduced an immunoreactive protein with a mobility similar toActA-CAAX (lane 5). Cells transfected with VSV-zyxin(381-559)-CAAX (the LIM domains) did not produce a signal thatcould be detected by anti-ActA3 or zyxin antibodies, althougha product of 25 kDa was detected with an anti-VSV antibody

(data not shown). These results confirm that anti-ActA anti-bodies react with zyxin protein, and that the common epitopeslie within amino acids 1-381 of zyxin, the region that sharessequence identity with ActA.

The domains of zyxin and ActA that share sequenceidentity aff ect the actin c ytoskeleton and cell shapein a similar mannerCells transiently transfected with DNA constructs encoding


Fig. 5. The domains common to ActA and to zyxinaffect the actin cytoskeleton and the shape of HeLaand CV-1 cells in a similar manner. HeLa cells (A-D) or CV-1 cells (E-J) were transfected with DNAconstructs encoding ActA(235-584) (A,B,E,F) orzyxin(1-381)-CAAX (C,D,G,H,I,J). At 30 hoursafter transfection cells were fixed, permeabilized,and doubly stained with anti-ActA3 antibody andwith phalloidin to mark F-actin. Right panels aremicrographs of immunofluorescence staining. Leftpanels are micrographs of the corresponding stainedF-actin. In C and D the two HeLa cells that producedzyxin(1-381)-CAAX have circumferential plasmamembrane blebs as observed in cells that producedActA(235-584). The CV-1 cell that producedzyxin(1-381)-CAAX (G,H) had no stress fibers asobserved in cells that produced ActA(235-584)-CAAX. (I,J) Plane of focus at the dorsal face ofcells. The cell which produced zyxin(1-381)-CAAXwas covered by numerous small structures. Note theco-distribution of F-actin and membrane-associatedzyxin(1-381)-CAAX. Bar, 10 µm.

ActA(1-584)-CAAX, wild-type zyxin, zyxin variants, or theActA-zyxin chimera were double labeled by rhodaminecoupled phalloidin to mark F-actin and by anti-ActA3 antibodyor by VSV monoclonal antibody. The labeling pattern of zyxinin transfected CV-1 cells that overproduced wild-type zyxinresembled the staining pattern of endogenous zyxin in untrans-fected cells (data not shown). In contrast, the addition of theK-ras (B) sequence to the carboxy-terminal end of zyxinvariants resulted in a homogenous distribution of the zyxin-label at the cell membrane, in a pattern that is typical of plasmamembrane-associated proteins (Fig. 5). The distribution of thelabel was restricted to the plasma membrane, as determined byconfocal microscopy (data not shown).

Production of zyxin(1-559)-CAAX and zyxin(1-381)-CAAXled to a reorganization of the actin cytoskeleton and changes incell shape in HeLa and CV-1 cell lines. Since the effects of thesetwo zyxin variants were indistinguishable from each other, onlycells that produced zyxin(1-381)-CAAX are shown (Fig. 5). InHeLa cells, production of ActA(235-584)-CAAX resulted in cir-cumferential plasma membrane blebbing as previously reported(A,B) (Friederich et al., 1995). Strikingly, expression of zyxin(1-381)-CAAX also produced circumferential plasma membrane

blebs (C,D). These plasma membrane blebs were distributed atthe dorsal surface of the cells, as visualized by confocalmicroscopy analysis (data not shown). In CV-1 cells, ActA(235-584)-CAAX disrupted actin stress fibers as previously reported(E,F). Production of zyxin(1-381)-CAAX in CV-1 cells alsodisrupted actin stress fibers resulting in a rhodamine phalloidinstaining pattern that resembled that of ActA(235-584)-CAAXtransfected cells (G,H). In addition to the modifications of theactin cytoskeleton observed at the ventral face of the cells, 20%of transfected CV-1 cells producing zyxin(1-381)-CAAX hadnumerous actin-rich membrane structures on their dorsalsurfaces that were not detected in non-transfected cells (I,J). Westress that cells transfected with proteinA-CAAX did not showany detectable changes in the actin cytoskeleton as has beenshown previously (Friederich et al., 1995). Likewise, productionof VSV-zyxin(381-559)-CAAX in HeLa cells or CV-1 cells didnot cause plasma membrane blebbing or stress fiber disruption.We noted, however, that in cells that had relatively intenseimmunofluorescence signals, the VSV-zyxin(381-559)-CAAXprotein appeared to be concentrated in large patches; frequently,by 48 hours after transfection the number of positive stainingcells was greatly reduced.


Fig. 6. An ActA(1-234)-zyxin(1-381)-CAAX chimera affects theactin cytoskeleton and the shape of HeLa and CV-1 cells in a similarmanner to wild-type ActA. HeLa cells (A-D) or CV-1 cells (E-H)were transfected with the DNA construct encoding ActA(1-584)-CAAX (A,B,E,F) or the chimera ActA(1-234)-zyxin(1-381)-CAAX(C,D,G,H). Cells were analyzed as described in Fig. 5. Right panelsare micrographs of immunofluorescence staining. Left panels aremicrographs of the corresponding stained F-actin. A plane of focus atventral face of cells is shown. HeLa cells which produced ActA(1-234)-zyxin(1-381)-CAAX (C,D) were well spread and madenumerous plasma membrane extensions as observed in cells thatproduced ActA(1-584)-CAAX (A,B). Note that in comparison withneighboring untransfected cells, the diffuse F-actin label is increasedin the CV-1 cell which produced ActA(1-234)-zyxin(1-381)-CAAX(G, H). Bars, 10 µm.

The amino-terminal domain of zyxin (1-381)functionall y replaces the medial/carbo xy-terminalpor tion of ActAOur results suggested that the portions of ActA and zyxin thatshare sequence similarity have similar functional activity. Wetested this hypothesis directly by replacing amino acids 235-584 of ActA with amino acids 1-381 of zyxin to make an ActA-zyxin chimera that is structurally similar to full length ActA(Fig. 4A). The N terminus of ActA (amino acids 1-234) isessential for the actin organizing activity of ActA, however,without the proline rich domain and C-terminal domain (aminoacids 235-584), its activity is limited (Friederich et al., 1995;Lasa et al., 1995; Smith et al., 1996). As described above,membrane targeted ActA (235-584) that lacks the N-terminaldomain, stimulates circumferential membrane blebbing in HeLacells. In contrast, HeLa cells that produce ActA(1-584)-CAAXdisplayed a flat shape, absence of stress fibers and numerousprominent cell membrane extensions as previously described(Fig. 6A,B) (Friederich et al., 1995). As predicted, HeLa cellsthat produced the ActA-zyxin-CAAX chimera also had a flatshape, absence of stress fibers and presented numerous lamellarplasma membrane extensions (C,D). In CV-1 cells transfectedwith ActA(1-584)-CAAX, the rhodamine-phalloidin stain washomogeneously distributed at the cell surface and co-distributedwith the staining of the protein, as shown in E,F. Redistributionof F-actin to the plasma membrane was observed by confocalmicroscopy analysis. A similar cell staining pattern wasobserved in CV-1 cells transfected with the ActA-zyxin-CAAXchimera (G,H). Moreover, as was the case for the ActA-CAAXproducing cells, the intensity of the F-actin label was higher ascompared with that of neighboring non-transfected cells.

Zyxin and ActA interact with the same pr oteins invitr oIt has been previously reported that zyxin and ActA bind invitro to members of the Ena/VASP protein family (Reinhard etal., 1995; Gertler et al., 1996). We tested if these interactionswere due to the domains common to ActA and zyxin. For thispurpose, we isolated VASP and Mena (mammalian Ena) byprofilin affinity chromatography as has been previouslydescribed. Total HeLa cell extracts (Fig. 7, lanes 1), andproteins isolated by profilin affinity chromatography from totalHeLa cell extracts (lanes 2) were separated by SDS-PAGE,transferred to nitrocellulose, and probed with either GST-ActA(235-584) protein or with GST-zyxin(1-348). Todetermine if ActA or zyxin bound to the immobilized proteins,the nitrocellulose was incubated with either anti-ActA3antibody or anti-zyxin antiserum. Development of themembranes after overlay and antibody incubation revealed apair of proteins of 90 kDa and 45 kDa present in the sampleobtained by profilin affinity chromatography. Both the 90 kDaand 45 kDa proteins were detected when either GST-ActA(235-584) or GST-zyxin(1-348) was used in the overlay.We noted that, as expected, both anti-ActA3 and anti-zyxinantibodies reacted with zyxin protein in total HeLa cell extracts(middle and right panels, lane 1); anti-zyxin antiserum alsoreacted with zyxin protein that bound to profilin-Sepharose(right panel, lane 2) as was previously reported (Gertler et al.,1996). No signals that corresponded to the 90 and 45 kDproteins were detected in the absence of overlay proteins (blankpanel) or with GST alone (data not shown).

a

The pair of proteins at 90 kDa co-migrated exactly withproteins recognized by anti-Mena antibodies (data not shown).Similarly, the 45 kDa protein co-migrated exactly with proteinrecognized by anti-VASP antibodies (data not shown). Theseresults support previous observations that ActA and zyxininteract with members of the Ena/VASP family in vitro. Fur-thermore, these results extend previous reports by demonstrat-ing that the portions of the proteins required for the interac-tions lie within the domains that are structurally andfunctionally similar. The other proteins in the unfractionatedHeLa cell extract that interact with the proline region of zyxinin this assay remain to be characterized further.


Fig. 7. The domains common to ActA and to zyxin interact in vitro withMena and VASP proteins. (A) Total HeLa cell extract (lanes 1) andproteins bound to profilin-Sepharose after incubation with HeLa cellextract (lanes 2) were analyzed by SDS-PAGE and immunoblotting. Themembranes were incubated with either blocking buffer alone (blank),blocking buffer with recombinant GST-ActA235-584 (1 µg/ml), orblocking buffer with recombinant chicken zyxin GST-zyxin1-348 (2µg/ml). Membranes were then developed with anti-ActA3 antibody (blankand GST-ActA235-584 panels) or with anti-zyxin serum (GST-zyxin1-348 panel). The top arrow indicates the position of Mena proteins and thelower arrow indicates the position of VASP as determined in separateexperiments. The only protein detected by Coomassie staining in lanes 2was actin whereas many proteins were detected by Coomassie staining inlanes 1 (not shown) confirming the specificity of the overlay assay. Zyxin was detected independently of the addition of overlay protein in totalcell extracts (lanes 1) and in lane 2 in the panel developed with anti-zyxin serum. The positions of molecular mass markers (kDa) are shown onthe left. (B,C) Membrane-associated zyxin(1-381)-CAAX causes a redistribution of VASP in CV-1 cells. CV-1 cells were transfected with aDNA construct encoding a variant of zyxin(1-381)-CAAX that was epitope-tagged with a peptide sequence derived from myc. Cells weredouble-labeled for immunofluorescence to detect VASP (B) and epitope-tagged zyxin(1-381)-CAAX (C). A plane of focus at the ventral face ofthe cells is shown. The VASP label was diffusely distributed at the surface in the cells which produced zyxin(1-381)-CAAX, whereas inneighboring untransfected cells the VASP label was distributed in focal adhesions as previously described. Bar, 10 µm.

Because of the ability of zyxin to interact with members ofthe Ena/VASP family, we examined whether production ofzyxin-CAAX variants by transient transfection would have aneffect on the subcellular distribution of VASP. Consistent withthe results of the binding in vitro, we found that VASP isdisplaced in CV-1 cells that were producing zyxin(1-381)-CAAX (Fig. 7B,C). VASP showed a diffuse staining pattern atthe plasma membrane and stronger staining where the twotransfected cell overlapped, similar to the staining pattern ofthe zyxin variant. In neighboring non-transfected cells, VASPstaining corresponded to focal adhesions as previouslydescribed (Reinhard et al., 1992).

DISCUSSION

ActA and zyxin ha ve common pr oper tiesWe used anti-ActA antibodies to identify and characterizeproteins that have properties similar to ActA. By this approach,we showed that anti-ActA1 and anti-ActA3 antibodies detechuman and simian zyxin. These antibodies stained actin richregions of the cell including focal adhesions and actin cortexof the lamellipodium, the same regions that are stained by anti-zyxin antibodies (Macalma et al., 1996). By two-dimensional

t

gel electrophoresis and immunoblotting techniques, anti-ActAantibodies recognized a series of closely spaced proteins ofapproximately 84 kDa that showed similar mobility to zyxin.Furthermore, zyxin variants produced by transient transfectionwere detected by anti-ActA antibodies. We conclude that L.monocytogenesActA and human zyxin have similar epitopes.

Many actin regulatory proteins that are important for con-trolling actin polymerization are associated with the plasmamembrane. The targeted transfection assay places proteins atthe inner plasma membrane by virtue of a CAAX motif and abasic stretch of amino acids derived from K-ras (B). By ectopi-cally placing proteins at this site, we are able to test their effectson the actin cytoskeleton. For example, we have previouslyidentified functionally distinct domains of ActA by thisapproach (Friederich et al., 1995). Here we show in two celllines that zyxin affects the actin cytoskeleton and cell shape ina similar manner to that of the related domains of ActA. Moreimportantly, we reconstituted an ActA effect in cells by linkingthe proline rich and C-terminal regions of zyxin with the N-terminal domain of ActA. Furthermore, we show that theregion common to both proteins is responsible for the observedinteractions with Ena/VASP family members.

Recently, L. monocytogenesactin comets in eucaryotic cellshave been used as a model system to describe the organization


of actin at sites along the plasma membrane in mammalian cells.This model predicted that a protein with ActA-like function ispresent in mammalian cells. The identification of zyxin by usinganti-ActA antibodies further supports the usefulness of L. mono-cytogenesas a model system for the study of actin organization.Analysis of ActA and zyxin by using highly purified antibodiessuggested that multiple epitopes are common to the two proteins.It is not clear if these common sequences are the result of genetransfer or convergent evolution. Within the Listeria genus, onlytwo of six Listeria species are pathogenic and are believed toeither have genes required for host cell infection or to have thesegenes expressed at sufficient levels for virulence (Sheehan et al.1994). The characterization of zyxin genes from many specieswill be required to clarify the evolutionary relationship of zyxinto ActA and IactA.

Proper ties of the zyxin pr otein famil yThe molecular architecture of zyxin is compatible with a roleas a docking site for the assembly of multimeric proteincomplexes. Zyxin displays an N-terminal domain that isenriched in proline repeats followed by three LIM domains.The proline-rich region harbors binding sites for α-actinin(Crawford et al., 1992), and as illustrated here, members of theEna/VASP family (Reinhard et al., 1995). A member of the dblprotein family, Vav, also interacts with the N-terminal domainof zyxin; the significance of this interaction remains to beresolved (Hobert et al., 1996). In addition, since single LIMdomains have been shown to function as protein binding inter-faces (Schmeichel and Beckerle, 1994; Wu and Gill, 1994), theLIM region of zyxin may serve to recruit multiple bindingpartners (Schmeichel and Beckerle, 1997).

Human zyxin is a member of a new family of proteins thatare related in their structural organization (Macalma et al.,1996; Zumbrunn and Trueb, 1996). Other members includechicken zyxin and the lipoma preferred partner (LPP) (Sadleret al., 1992; Petit et al., 1996). The number of different zyxin-like proteins present in one cell type is not yet known. Anti-zyxin antibodies and anti-ActA antibodies detected severalpoorly resolved proteins in fibroblasts by two-dimensional gelelectrophoresis (Beckerle, 1986; Crawford and Beckerle,1991), however, zyxin is known to be a phosphoprotein in vivo(Crawford and Beckerle, 1991) and it is likely that at least someof the zyxin isoelectric point variants are the result of differ-ential phosphorylation.

Zyxin and ActA ha ve a common domain that is ric hin pr olinesThe N-terminal domain of zyxin (amino acids 1-381) and thmedial/C-terminal portion of ActA (amino acids 235-584) arethe regions that contain proline and glutamate repeats(Macalma et al., 1996; Higley and Way, 1997). Mapping ofanti-ActA antibodies revealed that some epitopes are foundoutside of the proline rich regions indicating that the structuralfeatures extend beyond what is obvious from sequence align-ments. The limited sequence similarity (23% identity) betweenthese regions of zyxin and ActA is reminiscent of the rela-tionship between ActA and IactA (Gouin et al., 1995; Kreft etal., 1995). Furthermore, even within the zyxin family ofproteins, human LPP is only 33% identical to human zyxin(Macalma et al., 1996; Petit et al., 1996). A high resolutionstructural comparison of ActA and zyxin will be required to

,

e

examine the three-dimensional features of these proteins andextend the data we present here.

The sequence alignment of ActA and zyxin indicates thatLIM domains are not present in ActA although three are foundin zyxin (Sadler et al., 1992; Macalma et al., 1996; Zumbrunnand Trueb, 1996). LIM domains are found in many proteins ofdiverse function (Freyd et al., 1990; Schmeichel and Beckerle,1994). These domains are characterized by the presence ofseveral cysteine residues that are required to co-ordinate zincions (Kosaet al., 1994; Perez-Alvarado et al., 1994). Directtesting showed that anti-ActA antibodies did not react withzyxin LIM domains. Given that LIM domains are not found inActA, we propose that the properties common to ActA andzyxin map to the N-terminal region of zyxin, as suggested byresults from the targeted transfection assay.

Sequence alignments highlight the fact that zyxin lacks anysimilarity to the N-terminal domain of the ActA protein. Wepreviously found that this region of ActA is essential for actinrecruitment but not sufficient for plasma membrane extension(Friederich et al., 1995). In addition, L. monocytogenesthatharbor deletions in the N-terminal domain of ActA assembleF-actin but fail to move in Caco-2 or PtK2 cells (Lasa et al.,1995). In cell free Xenopusextracts, however, L. monocyto-genesthat express only the N-terminal 234 amino acids ofActA can assemble F-actin and move, albeit at a rate threefoldless than for wild-type bacteria (Lasa et al., 1997). Theseexperiments suggest that the N-terminal domain of ActA isessential for the initiation of actin polymerization.

Zyxin’ s N-terminal domain has similar pr oper ties tothat of ActAThe organization of the actin cytoskeleton and cell shape of twocell lines were affected in a similar manner by the amino-terminal domain of zyxin and by the medial and carboxy-terminal domain of ActA. The plasma membrane associated N-terminal 381 amino acids of human zyxin strongly reduced thenumber and the size of stress fibers in CV-1 cells, as previouslyreported for the structurally similar region of ActA. In HeLacells this variant caused a dramatic membrane blebbing also asreported for the related domain from ActA. Cell surface blebshave been observed in a number of cell types and are thought tobe associated with weakening of the actin cortical cytoskeleton(Cunningham, 1995). The effects of the N-terminal domain ofzyxin were indistinguishable from those of full length zyxin (i.e.zyxin including LIM domains) by this assay suggesting that theActA-lik e properties lie within the N-terminal region. Theeffects of LIM domains alone were difficult to interpret becausethe membrane-targeted protein formed aggregates whenproduced in transient transfection assays.

Since the effects on actin cytoskeleton and cell shape causedby the N-terminal region of zyxin were similar to those of themedial and C-terminal region of ActA, we tested a chimera thatencoded the N-terminal domain of ActA in fusion with the N-terminal domain of zyxin. Upon transfection of either CV-1 orHeLa cells, this construct fully restored an ActA phenotype. InHeLa cells, surface blebbing was absent and, instead, strikinglamellar shaped plasma membrane extensions were observed.In both cell lines, F-actin was redistributed and recruited to theplasma membrane. These results suggest that human zyxin hasnot only common structural features with ActA but alsoexhibits similar functional properties.


ActAct

ActAct

ActAct

ActAct

ActAct

Act

P

Act

vaspAct Act

Act

Act

Pvasp

P

P

P

P

P

P

X Y

PAct

P

ActAct

ActAct

Act

A BPlasma membrane Plasma membrane

Low rate of actin turnover High rate of actin turnover

ActAct

ActAct

Act

Pvasp

Pvasp

P

P

X Y

P

P

P

P

ActAct

ActAct

ActAct

ActAct

Act

Act ActAct

ActAct

ActAct

ActActAct

ActAct

ActAct

Act

Act

Act

Act

Act

ActAct

Act

Act

Act

Act

Act

Act

Act

Act

zyxin

zyxin

zyxin

zyxin

Fig. 8. A model to describe zyxin function. We present a model to propose a function for zyxin based upon data presented here and elsewhere(see Introduction and Discussion for citations). Zyxin, VASP, profilin (P) and actin (Act) proteins are labeled. We propose that zyxin can bind totwo different classes of proteins. One class of zyxin interacting proteins is represented by VASP, which binds profilactin complexes. Wepropose that zyxin interacts with a second class of proteins (or protein complexes) represented here by X and Y. These proteins are likely to belocated at the plasma membrane and have basal level of actin organizing activity. (A) Zyxin is part of a complex that contains VASP, profilinand actin. The molar ratio of each component is not considered here, although the intracellular profilactin concentration is in excess over zyxin.Without being bound to an actin organizing site, the zyxin complex has no net effect on the actin cytoskeleton. At the plasma membranedifferent proteins that organize F-actin structures are in a state of low actin turnover. (B) Zyxin is part of a complex that contains VASP, profilinand actin. This complex can associate with different F-actin organizing sites at the plasma membrane via different binding sites on zyxin. Cellstaining experiments show that zyxin is located at different actin rich sites in cells. We propose that actin could be transferred from profilactinto F-actin, and new profilactin complexes would be directed to vacant VASP binding sites. We propose that the increased access to actincontributes to a relatively high rate of actin turnover at the plasma membrane. The interaction of zyxin-VASP complexes with proteins at theplasma membrane may be regulated by signal transduction mechanisms that are known to affect the actin cytoskeleton. Depending on the typeof signal pathway activated, zyxin may bind to only one actin organizing center at a time (for example, either X or Y), although both of therepresentative sites are shown to be occupied in this panel. The ability of zyxin-VASP complex to interact specifically with one site or anotherwould result in spatially or temporally controlled changes in the actin cytoskeleton.

If zyxin and ActA have similar properties, it would be likelythat they interact with the same proteins in cells. We looked forzyxin and ActA interacting proteins by using an overlay assay.As previously described, zyxin and ActA interact with membersof the Ena/VASP family (Reinhard et al., 1995; Gertler et al.,1996). We confirmed and extended these observations bydemonstrating that common domains of ActA and zyxin weresufficient for this interaction. In order to clearly detect theseinteractions, it was necessary to enrich for VASP and Menaproteins by profilin affinity chromatography. Relative to actinand to proteins present in total HeLa cell extract, however,VASP and Mena were not visible by Coomassie staining, indi-cating that in this type of assay the zyxin and ActA interactionwas specific. In a subsequent assay, VASP was displaced from

focal adhesions by ectopic localization of the N-terminaldomain of zyxin at the plasma membrane. VASP is a 45 kDaprotein that was originally identified as a major substrate forcAMP and cGMP dependent protein kinases in platelet activa-tion signaling pathways (Halbrügge and Walter, 1989). It isfound in actin rich structures such as focal adhesions, cell-cellcontacts and at the plasma membrane (Reinhard et al., 1992).VASP has the interesting property of binding to profilin and tovinculin in vitro (Reinhard et al., 1995a, 1996; Brindle et al.,1996). The ectopic placement of zyxin is a powerful techniqueto analyze the composition of protein complexes in actin richstructures. Having displaced VASP, we are currently testing ifother proteins are lost from focal adhesions, and the conse-quences of this displacement.

1904

te

R. M. Golsteyn and others

Accelerating actin d ynamics with zyxin: a modelBy transfection of zyxin variants we observed disruption ofstress fibers, plasma membrane blebbing, and formation ofactin rich structures on the dorsal surfaces of cells. Despite thefact that these effects are induced by protein overproductionthis approach has been useful in identifying functional domainin ActA and in zyxin. Furthermore, by preparing zyxin mutantswe have used this approach to identify subdomains withinzyxin that are responsible for specific cytoskeletal effects (B.Drees and M. C. Beckerle; R. M. Golsteyn and E. Friederich;unpublished results). These effects are consistent with theproposal that zyxin plays a role in the organization of differentactin structures in cells.

We propose a model in which zyxin is part of a multiproteincomplex that has the capacity to accelerate the activity of otheractin interacting proteins (Fig. 8). The zyxin complex wouldnot have a capacity to make new actin cytoskeleton structureson its own; rather, it would contribute to the actin organizingactivity of other proteins. This model predicts that zyxininteracts with at least two distinct classes of proteins. One classof proteins would be responsible for the recruitment of actinmonomers. VASP would be an example of a member of thisclass of proteins. The interaction with VASP and othermembers of the Ena/VASP protein family would lead to theformation of a complex that contains profilactin. Recentevidence suggests that vinculin binds VASP in vitro (Brindleet al., 1996; Reinhard et al., 1996), it would be interesting toknow if vinculin and zyxin has overlapping roles in organizingthe actin cytoskeleton in cells.

The second class of zyxin interacting proteins would beexpected to have the ability to create new F-actin structures,for example, by having actin nucleating activity or F-actinuncapping activity. The role of zyxin would be to link thesetwo classes of proteins in order to deliver actin monomers tosites of actin organization. Zyxin could deliver actin monomersto a variety of sites which may have distinctly different activ-ities. In each case, the transient binding of the zyxin complexwould enhance an actin organizing activity by increasing thelocal actin monomer concentration.

Data in support of this model come from analysis of ActAand zyxin. The actin organizing activity of ActA lies within itsN-terminal 234 amino acids (Friederich et al., 1995; Pistor etal., 1995; Lasa and Cossart, 1996; Smith et al., 1996; Lasa eal., 1997). The medial/C-terminal domain of ActA enhancesthe activity present in the N-terminal domain. The N-terminaldomain of zyxin has a similar enhancing activity as suggestedby full restoration of ActA-like activity by the zyxin-ActAchimera. The localization of zyxin in cells is consistent with arole in the regulation of actin structure. In eucaryotic cells,whose requirements for actin organization are spatially andtemporally complex, a protein assemblage with an accelerat-ing activity component might be more versatile than the‘covalent’ attachment of an accelerating activity as foundwithin ActA. The recent advances in our understanding of therole of the family of small GTPases in the organization of theactin cytoskeleton suggest possible signals that may direct azyxin complex to a particular site within the cell (Nobes andHall, 1995).

We thank Antoine Durrbach and Fatima El Marjou for help inpreparing ActA fusion proteins and rabbit immunizations; Iñigo Lasa,

s

t

Edith Gouin and Pascale Cossart for Listeria ivanovii extracts and dis-cussions; Jean Salamero for help with confocal microscopy; BrigitteBauvois, Ariane Rochat and Christian Gachet for supplying cells andcell lines; Steve Almo for the profilin expression plasmid; and UlrichWalter, Frank Gertler, Thomas Kreis, and Victor Kotelianski for anti-bodies. We also thank Daniel Louvard for encouragement andvaluable discussions. This work was supported by grants from theAssociation pour la Recherche sur le Cancer (ARC), the LigueNationale Francaise contre le Cancer, contract CNRS ‘Biologie Cel-lulaire: Du Normal au Pathologie,’ the Institut Curie, the NationalInstitutes of Health (GM50877) and the Huntsman Cancer Institu.M.C.B. is a recipient of a Faculty Research Award from the AmericanCancer Society. R.M.G. is a recipient of fellowships from ARC andfrom the Human Frontiers Science Program.

REFERENCES

Barkalow, K., Witk e, W., Kwiatkowski, D. J. and Hartwig, J. H. (1996).Coordinated regulation of platelet actin filament barbed ends by gelsolin andcapping protein. J. Cell Biol. 134, 389-399.

Beckerle, M. C. (1986). Identification of a new protein localized at sites of cell-substrate adhesion. J. Cell Biol. 103, 1679-1687.

Br indle, N. P., Holt, M. R., Davies, J. E., Price, C. J. and Critchely, D. R.(1996). The focal-adhesion vasodilator-stimulated phosphoprotein (VASP)binds to the proline-rich domain in vinculin. Biochem. J. 318, 753-757.

Burnette, W. N. (1981). ‘Western blotting’: electrophoretic transfer of proteinsfrom sodium dodecyl sulfate-polyacrylamide gels to unmodifiednitrocellulose and radiographic detection with antibody and radioiodinatedprotein A. Anal. Biochem. 112, 195-203.

Cano, M. L., Lauffenburger, D. A. and Zigmond, S. H. (1991). Kineticanalysis of F-actin depolymerization in polymorphonuclear leukocytelysates indicates that chemoattractant stimulation increases actin filamentlength distribution. J. Cell Biol. 115, 677-687.

Chakraborty, T., Ebel, F., Domann, E., Niebuhr, K., Gerstel, B., Pistor, S.,Temm-Grove, C. J., Jockusch, B. M., Reinhard, M., Walter, U. andWehland, J. (1995). A focal adhesion factor directly linking intracellularlymotile Listeria monocytogenes and Listeria ivanovii to the actin-basedcytoskeleton of mammalian cells. EMBO J. 14, 1314-1321.

Collins, S. J., Gallo, R. C. and Gallagher, R. E. (1977). Continuous growthand differentiation of human myeloid leukaemic cells in suspension culture.Nature 270, 347-349.

Crawford, A. W. and Beckerle, M. C. (1991). Purification andcharacterization of zyxin, an 82,000-Dalton component of adherensjunctions. J. Biol. Chem. 266, 5847-5853.

Crawford, A. W., Michelsen, J. W. and Beckerle, M. C. (1992). Aninteraction between zyxin and alpha-actinin. J. Cell Biol. 116, 1381-1393.

Cudmore, S., Cossart, P., Grif fiths, G. and Way, M. (1995). Actin-basedmotility of vaccinia virus. Nature 378, 636-638.

Cunningham, C. C. (1995). Actin polymerization and intracellular solventflow in cell surface blebbing. J. Cell Biol. 129, 1589-1599.

Dabir i, G. A., Sanger, J. M., Portnoy, D. A. and Southwick, F. S. (1990).Listeria monocytogenes moves rapidly through the host-cell cytoplasm byinducing directional actin assembly. Proc. Nat. Acad. Sci. USA 87, 6068-6072.

Domann, E., Wehland, J., Rohde, M., Pistor, S., Hartl, M., Goebel, W.,Leimeister-Wächter, M., Wuenschner, M. and Chakraborty, T. (1992). Anovel bacterial virulence gene in Listeria monocytogenes required for hostcell microfilament interaction with homology to the proline-rich region ofvinculin. EMBO J. 11, 1981-1990.

Evan, G. I., Lewis, G., Ramsay, G. and Bishop, J. M. (1985). Isolation ofmonoclonal antibodies specific for human c-myc proto-oncogene product.Mol. Cell. Biol. 5, 3610-3616.

Fedorov, A. A., Pollard, T. D. and Almo, S. C. (1994). Purification,characterization and crystalization of human platelet profilin expressed inEscherichia coli. J. Mol. Biol. 241, 480-482.

Forscher, P. and Smith, S. J. (1988). Actions of cytochalasins on theorganization of actin filaments and microtubules in a neuronal growth cone.J. Cell Biol. 107, 1505-1516.

Freyd, G., Kim, S. K. and Horvitz, H. R. (1990). Novel cysteine-rich motifand homeodomain in the product of the Caenorhabditis elegans cell lineagegene lin 11. Nature 344, 876-879.


Fr iederich, E., Huet, C., Arpin, M. and Louvard, D. (1989). Villin inducesmicrovilli g rowth and actin redistribution in transfected fibroblasts. Cell 59,461-475.

Fr iederich, E., Gouin, E., Hellio, R., Kocks, C., Cossart, P. and Louvard, D.(1995). Targeting of Listeria monocytogenes ActA protein to the plasmamembrane as a tool to dissect both actin-based cell morphogenesis and ActAfunction. EMBO J. 14, 2731-2744.

Gerstel, B., Gröbe, L., Pistor, S., Chakraborty, T. and Wehland, J. (1996).The ActA polypeptides of Listeria ivanovii and Listeria monocytogenesharbor related binding sites for host microfilament proteins. Infect. Immun.64, 1929-1936.

Gertler, F. B., Niebuhr, K., Reinhard, M., Wehland, J. and Soriano, P.(1996). Mena, a relative of VASP and Drosophila enabled, is implicated inthe control of microfilament dynamics. Cell 87, 227-239.

Gossen, M. and Bujard, H. (1992). Tight control of gene expression inmammalian cells by tetracycline-responsive promoters. Proc. Nat. Acad. Sci.USA 89, 5547-5551.

Gouin, E., Dehoux, P., Mengaud, J., Kocks, C. and Cossart, P. (1995). iactAof Listeria ivanovii, although distantly related to Listeria monocytogenesactA, restores actin tail formation in an L. monocytogenes actA mutant.Infect. Immun. 63, 2729-2737.

Halbrügge, M. and Walter, U. (1989). Purification of a vasodilator-regulatedphosphoprotein from human platelets. Eur. J. Biochem. 185, 41-51.

Hancock, J. F., Paterson, H. and Marshall, C. J. (1990). A polybasic domainor palmitoylation is required in addition to the CAAX motif to localizep21ras to the plasma membrane. Cell 63, 133-139.

Hancock, J. F., Cadwallader, K., Paterson, H. and Marshall, C. J. (1991). ACAAX or a CAAL motif and a second signal are sufficient for plasmamembrane targeting of ras proteins. EMBO J. 10, 4033-4039.

Har low, E. and Lane, D. (1988). Antibodies: a Laboratory Manual. ColdSpring Harbor Laboratory Press, Cold Spring Harbor.

Hartwig, J. H., Bokoch, G. M., Carpenter, C. L., Janmey, P. A., Taylor, L.A., Toker, A. and Stossel, T. P. (1995). Thrombin receptor ligation andactivated Rac uncap actin filament barbed ends through phosphoinositidesynthesis in permeabilized human platelets. Cell 82, 643-653.

Higley, S. and Way, M. (1997). Actin and cell pathogenesis. Curr. Opin. CellBiol. 9, 62-69.

Hobert, O., Schilling , J. W., Beckerle, M. C., Ullr ich, A. and Jallal, B.(1996). SH3 domain-dependent interaction of the proto-oncogene productVav with the focal contact protein zyxin. Oncogene 12, 1577-1581.

Kocks, C., Gouin, E., Tabouret, M., Berche, P., Ohayon, H. and Cossart, P.(1992). L. monocytogenes-induced actin assembly requires the actA geneproduct, a surface protein. Cell 68, 521-531.

Kocks, C., Marchand, J. B., Gouin, G., d’Hauteville, H., Sansonetti, P. J.,Carlier, M. F. and Cossart, P. (1995). The unrelated surface proteins ActAof Listeria monocytogenes and IcsA of Shigella flexneri are sufficient toconfer actin-based motility on Listeria innocula and Escherichia colirespectively. Mol. Microbiol. 18, 413-423.

Kosa, J. L., Michelsen, J. W., Louis, H. A., Olsen, J. I., Davis, D. R.,Beckerle, M. C. and Winge, D. R. (1994). Common metal ion coordinationin LIM domain proteins. Biochemistry 33, 468-477.

Kr eft, J., Dumbsky, M. and Theiss, S. (1995). The actin polymerizationprotein from Listeria ivanovii is a large repeat protein which shows onlylimited amino-acid sequence homology to ActA from Listeriamonocytogenes. FEMS Microbiol. Lett. 126, 113-122.

Laemmli, U. K. (1970). Cleavage of structural proteins during the assembly ofthe head of bacteriophage T4. Nature 227, 680-685.

Lasa, I., David, V., Gouin, E., Marchand, J. P. and Cossart, P. (1995). Theamino-terminal part of ActA is critical for the actin-based motility of Listeriamonocytogenes; the central proline-rich region acts as a stimulator. Mol.Microbiol. 18, 425-426.

Lasa, I. and Cossart, P. (1996). Actin-based bacterial motility: Towards adefinition of the minimal requirements. Trends Cell Biol. 6, 109-114.

Lasa, I., Gouin, E., Goethals, M., Vancompernolle, K., David, V.,Vandekerckhove, J. and Cossart, P. (1997). Identification of two regions inthe amino terminal domain of ActA involved in the actin comet tail formationby Listeria monocytogenes. EMBO J. 16, 1531-1540.

Lauf fenburger, D. A. and Horwitz, A. F. (1996). Cell migration: A physicallyintegrated molecular process. Cell 84, 359-369.

Macalma, T., Otte, J., Hensler, M. E., Bockholt, M., Louis, H. A., Kalf f-Suske, M., Grzeschik, K., von der Ahe, D. and Beckerle, M. C. (1996).Molecular characterization of human zyxin. J. Biol. Chem. 271, 31470-31478.

Matsuoka, Y., Moore, G. E., Yagi, Y. and Pressman, D. (1967). Production of

free light chains of immunoglobulin by a hematopoietic cell line derivedfrom a patient with multiple myeloma. Proc. Soc. Exp. Biol. Med. 125, 1246-1250.

Matthias, P. D., Renkawitz, R., Grez, M. and Schutz, G. (1982). Transientexpression of the chicken lysozyme gene after transfer into human cells.EMBO J. 1, 1207-1212.

Nobes, C. D. and Hall, A. (1995). Rho, Rac, and Cdc42 GTPases regulate theassembly of multimolecular focal complexes associated with actin stressfibers, lamellipodia, and filopodia. Cell 81, 53-62.

O’Farrell, P. Z., Goodman, H. M. and O’Farrell, P. H. (1977). Highresolution two-dimensional electrophoresis of basic as well as acidicproteins. Cell 12, 1133-1142.

Okabe, S. and Hirokawa, N. (1989). Incorporation and turnover of biotin-labeled actin microinjected into fibroblastic cells: An immunoelectronmicroscopic study. J. Cell Biol. 109, 1581-1595.

Pantaloni, D. and Carlier, M.-F. (1993). How profilin promotes actin filamentassembly in the presence of thymosinß4. Cell 75, 1007-1014.

Perez-Alvarado, G. C., Miles, C., Michelse, J. W., Louis, H. A., Winge, D.R., Beckerle, M. C. and Summers, M. F. (1994). Structure of the carboxy-terminal LIM domain from the cysteine rich protein CRP. Nature Struct. Biol.1, 388-398.

Petit, M. M., Mols, R., Schoenmakers, E. F., Mandahl, N. and Van De Ven,W. J. (1996). LPP, the preferred fusion partner gene of HMGIC in lipomas, isa novel member of the LIM protein gene family. Genomic 36, 118-129.

Pistor, S., Chakraborty, T., Walter, U. and Wehland, J. (1995). The bacterialactin nucleator protein ActA of Listeria monocytogenes contains multiplebinding sites for host microfilament proteins. Curr. Biol. 5, 517-525.

Reinhard, M., Halbrügge, M., Scheer, U., Wiegand, C., Jockusch, B. M.and Walter, U. (1992). The 46/50 kDa phosphoprotein VASP purified fromhuman platelets is a novel protein associated with actin filaments and focalcontacts. EMBO J. 11, 2063-2070.

Reinhard, M., Giehl, C., Abel, K., Haffner, C., Jarchau, T., Hoppe, V.,Jockusch, B. M. and Walter, U. (1995a). The proline-rich focal adhesionand microfilament protein VASP is a ligand for profilins. EMBO J. 14, 1583-1589.

Reinhard, M., Jouvenal, K., Tr ipier, D. and Walter, U. (1995b).Identification, purification, and characterization of a zyxin-related proteinthat binds the focal adhesion and microfilament protein VASP (vasodilator-stimulated phosphoprotein). Proc. Nat. Acad. Sci. USA 92, 7956-7960.

Reinhard, M., Rudiger, M., Jockusch, B. M. and Walter, U. (1996). VASPinteraction with vinculin: a recurring theme of interactions with proline-richmotifs. FEBS Let. 399, 103-107.

Robine, S., Sahuquillo-Merino, C., Louvard, D. and Pringault, E. (1993).Regulatory sequences on the human villin gene trigger the expression of areporter gene in a differentiating HT29 intestinal cell line. J. Biol. Chem. 268,11426-11434.

Rogel-Gaillard, C., Breitburd, F. and Orth, G. (1992). Humanpapillomavirus type 1 E4 proteins differing by their N-terminal ends havedistinct cellular localizations when transiently expressed in vitro. J. Virol. 66,816-823.

Sadler, I., Crawford, A. W., Michelsen, J. W. and Beckerle, M. C. (1992).Zyxin and cCRP: Two interactive LIM domain proteins associated with thecytoskeleton. J. Cell Biol. 119, 1573-1587.

Sambrook, J., Fr itsch, E. F. and Maniatis, T. (1989). Molecular Cloning: aLaboratory Manual. Cold Spring Harbor Laboratory Press, Cold SpringHarbor, NY.

Schafer, D. A., Jennings, P. B. and Cooper, J. A. (1996). Dynamics of cappingprotein and actin assembly in vitro: Uncapping barbed ends bypolyphosphoinositides. J. Cell Biol. 135, 169-179.

Schmeichel, K. L. and Beckerle, M. C. (1994). The LIM domain is a modularprotein-binding interface. Cell 79, 211-219.

Schmeichel, K. L. and Beckerle, M. C. (1997). Molecular dissection of a LIMdomain. Mol. Biol. Cell 8, 219-230.

Schmidt-Zachmann, M. S. and Nigg, E. A. (1993). Protein localization to thenucleolus: a search for targeting domains in nucleolin. J. Cell Sci. 105, 799-806.

Sheehan, B., Kocks, C., Dramsi, S., Gouin, E., Klarsfeld, A. D., Mengaud, J.and Cossart, P. (1994). Molecular and genetic determinants of the Listeriamonocytogenes infectious process. In Current Topics in Microbiology andImmunology: Bacterial Pathogenesis of Plants and Animals(ed. J. L. Dangl),pp. 187-216. Springer-Verlag, Heidelberg.

Small, J. V., Isenberg, G. and Celis, J. E. (1978). Polarity of actin at theleading edge of cultured cells. Nature 272, 638-639.

Smith, D. B. and Johnson, K. S. (1988). Single-step purification of


polypeptides expressed in Escherichia coli as fusions with glutathione S-transferase. Gene 67, 31-40.

Smith, G. A., Portnoy, D. A. and Theriot, J. A. (1995). Asymmetricdistribution of the Listeria monocytogenes ActA protein is required andsufficient to direct actin-based motility. Mol. Microbiol. 17, 945-951.

Smith, G. A., Theriot, J. A. and Portnoy, D. A. (1996). The tandem repeatdomain in the Listeria monocytogenes ActA protein controls the rate of actin-based motility, the percentage of moving bacteria, and the localization ofvasodilator-stimulated phosphoprotein and profilin. J. Cell Biol. 135, 647-660.

Southwick, F. S. and Purich, D. L. (1994). Arrest of Listeria movement in hostcells by a bacteria actA analogue: Implications for actin-based motility. Proc.Nat. Acad. Sci. USA 91, 5168-5172.

Stossel, T. P. (1993). On the crawling of animal cells. Science 260, 1086-1094.Theriot, J. A. (1995). The cell biology of infection by intracellular bacterial

pathogens. In Annual Review of Cell and Developmental Biology (ed. J. A.Spudich),pp. 213-239. Annual Reviews Inc., Palo Alto, Ca.

Theriot, J. A. and Mitchison, T. J. (1991). Actin microfilament dynamics inlocomoting cells. Nature 352, 126-131.

Tilney, L. G. and Portnoy, D. A. (1989). Actin filaments and the growth,movement, and spread of the intracellular bacterial parasite, Listeriamonocytogenes. J. Cell Biol. 109, 1597-1608.

Tilney, L. G., Connelly, P. S. and Portnoy, D. A. (1990). Actin filamentnucleation by the bacterial pathogen Listeria monocytogenes. J. Cell Biol.111, 2979-2988.

Tilney, L. G., DeRosier, D. J. and Tilney, M. S. (1992a). How Listeria exploitshost cell actin to form its own cytoskeleton. I. Formation of a tail and how thattail might be involved in movement. J. Cell Biol. 118, 71-81.

Tilney, L. G., DeRosier, D. J., Weber, A. and Tilney, M. S. (1992b). HowListeria exploits host cell actin to form its own cytoskeleton. II. Nucleation,actin filament polarity, filament assembly, and evidence for a pointed endcapper. J. Cell Biol. 118, 83-93.

Wang, Y.-L. (1985). Exchange of actin subunits at the leading edge of livingfibroblasts: Possible role of treadmilling. J. Cell Biol. 101, 597-602.

Welch, M. D., Iwamatsu, A. and Mitchison, T. J. (1997). Actinpolymerization is induced by Arp2/3 protein complex at the surface ofListeria monocytogenes. Nature 385, 265-269.

Wu, R. Y. and Gill, G. N. (1994). LIM domain recognition of a tyrosine-containing tight turn. J. Biol. Chem. 269, 25085-25090.

Zumbr unn, J. and Trueb, B. (1996). A zyxin-related protein whose synthesisis reduced in virally transformed fibroblasts. Eur. J. Biochem. 241, 657-663.

(Received 30 May 1997 – Accepted 13 June 1997)

structural and functional similarities between the human ... › content › joces › 110 › 16...

Documents